{
  "$type": "site.standard.document",
  "bskyPostRef": {
    "cid": "bafyreidqawoofgursm3l6y7ibja3ven57klljwrl4ccqeblbvbsb7rx6ui",
    "uri": "at://did:plc:lk3jfj3zq4k4wxnk474axylu/app.bsky.feed.post/3mmbnwwxm6g22"
  },
  "path": "/t/codex-credits-are-draining-while-idle-not-actively-used/1380250?page=2#post_39",
  "publishedAt": "2026-05-20T10:00:17.000Z",
  "site": "https://community.openai.com",
  "tags": [
    "Codex rate card",
    "Codex rate limits discussion thread"
  ],
  "textContent": "Rafa3’s suggestion is a great example and worth trying first.\n\nOne thing that can catch people off guard is that memory-related background work can still use credits even when Codex looks idle. So if you want to reduce that kind of usage, turning memory generation off can help.\n\nAlso, credit usage can vary a lot more now based on the model, speed mode, and the amount of work Codex is doing in a session. One prompt can cause a pretty noticeable drop if the task is complex, or if Codex has to scan, review, or analyze a large codebase, since that drives up token usage pretty quickly.\n\nIf you want the pricing details and a better sense of how usage is calculated now, this is the best reference: Codex rate card.\n\nWe’re going to close this thread out here, but please keep the discussion going in the main Codex limits thread so everything stays in one place: Codex rate limits discussion thread.\n\nThank you all again for engaging in this community. We really appreciate all this feedback.",
  "title": "Codex credits are draining while idle / not actively used"
}